Deprecate keras_model_custom()

Author: t-kalinowski

NEWS.md

@@ -2,28 +2,31 @@
 
 - Breaking change: The semantics of passing a named list to `keras_model()` have changed.
 
-  Previously, `keras_model()` would `unname()` supplied `inputs` and `outputs`.   
-  Then, if a named list was passed to subsequent 
-  `fit()`/`evaluate()`/`call()`/`predict()` invocations, 
-  matching of `x` and `y` was done to the model's `tensor$name`'s. 
+  Previously, `keras_model()` would `unname()` supplied `inputs` and `outputs`.
+  Then, if a named list was passed to subsequent
+  `fit()`/`evaluate()`/`call()`/`predict()` invocations,
+  matching of `x` and `y` was done to the model's `tensor$name`'s.
   Now, matching is done to `names()` of `inputs` and/or `outputs`.
   Call `unname()` on `inputs` and `outputs` to restore the old behavior, e.g.:
     ```keras_model(unname(inputs), unname(outputs))```
 
-  `keras_model()` can now accept a named list for multi-input and/or multi-output 
+  `keras_model()` can now accept a named list for multi-input and/or multi-output
   models. (Requires Tensorflow >= 2.4). The named list is converted to a `dict` in python.
-  
+
   If `inputs` is a named list:
-    - `call()`, `fit()`, `evaluate()`, and `predict()` methods can also 
-    accept a named list for `x`, with names matching to the 
+    - `call()`, `fit()`, `evaluate()`, and `predict()` methods can also
+    accept a named list for `x`, with names matching to the
     names of `inputs` when the model was constructed.
     Positional matching of `x` is still also supported (requires python 3.7+).
-      
+
   If `outputs` is a named list:
     - `fit()` and `evaluate()` methods can *only*
-    accept a named list for `y`, with names matching to the 
+    accept a named list for `y`, with names matching to the
     names of `outputs` when the model was constructed.
-      
+
+- `keras_model_custom()` is soft deprecated. Please define custom models by
+  subclassing `keras$Model` directly using `%py_class%` or `R6::R6Class()`.
+
 - Fixed warning issued by `k_random_binomial()`.
 - Fixed error raised when `k_random_binomial()` was passed a non-floating dtype.
 - Added `k_random_bernouli()` as an alias for `k_random_binomial()`.

R/model-custom.R

@@ -1,14 +1,18 @@
 
-#' Create a Keras custom model
+#' (Deprecated) Create a Keras custom model
+#'
+#' `keras_model_custom()` is soft-deprecated. Please define custom models by
+#' subclassing `keras$Model` directly using [`%py_class%`] or [`R6::R6Class()`].
 #'
 #' @param model_fn Function that returns an R custom model
 #' @param name Optional name for model
 #'
 #' @return A Keras model
 #'
-#' @details For documentation on using custom models,
-#'   see <https://keras.rstudio.com/articles/custom_models.html>.
+#' @details For documentation on using custom models, see
+#'   <https://keras.rstudio.com/articles/custom_models.html>.
 #'
+#' @keywords internal
 #' @export
 keras_model_custom <- function(model_fn, name = NULL) {